STUDIES IN YOUTH ALCOHOL CONSUMPTION AND PREVENTION

Transcription

1 From the Department of Public Health Sciences Division of Social Medicine Karolinska Institutet, Stockholm, Sweden STUDIES IN YOUTH ALCOHOL CONSUMPTION AND PREVENTION Mats Hallgren Stockholm 2012

2 All previously published papers were reproduced with permission from the publisher. Published by Karolinska Institutet. Printed by US AB printing, Stockholm. Mats Hallgren, 2012 ISBN

3 ABSTRACT Background: Alcohol remains the drug of choice for most young people and is responsible for a sizable proportion of deaths and injuries every year. In Sweden, total consumption and rates of heavy episodic drinking have reduced over the past ten years. At the same time, the number of adolescents admitted to hospital as a consequence of their drinking has risen. This unexpected trend warrants explanation with empirical research. The recent increase in serious alcohol-related harms also suggests there is more to learn about what works in prevention, including the effects of communitybased approaches and targeted brief interventions. Objectives: The thesis has two main objectives; first, to describe recent trends in alcohol consumption among Swedish youth, with a particular focus on polarisation effects (Study I). The second objective is to examine the effects of various alcohol prevention strategies targeting young people, and what can be learnt from these interventions (Studies II-IV). Methods: Study I (polarized youth drinking) uses repeated cross-sectional self-report data from the Stockholm Student Survey to explore changes in alcohol consumption and risk factors associated with heavy drinking among year 9 and year 11 students in Stockholm between 2000 and Changes in the dispersion of consumption over time are reported. Study II also uses cross-sectional data to examine the effects of a comprehensive alcohol prevention trial targeting young people in 12 communities in Sweden between 2003 and Studies III and IV assess the effectiveness of a brief health education program on consumption and attitudes towards alcohol in high schools and the Swedish military, with assessments taken at 5 and 20 month follow-up. All participants were aged between 15 and 20 years. Results: Findings indicate that a polarization in youth drinking is a likely explanation for the recent divergence between alcohol consumption and serious alcohol-related harms among youth. We found significant increases in the dispersion of consumption over time, indicating more heavy drinkers in the tail end of the drinking distribution. Most adolescent in Stockholm continue to drink less or abstain from alcohol completely, but a minority continue to drink more alcohol. Results concerning the relationship between heavy drinking and risk factors were inconclusive. We found no significant improvements in six trial communities compared to six control communities following a four year multi-component community intervention primarily targeting young people, although adults in the trial communities developed more restrictive attitudes towards the supply of alcohol. The Prime for Life brief health education program did not lead to significant improvements in alcohol use or attitudes towards alcohol in either high school students or military conscripts. Conclusion: We suggest that ongoing social changes could be affecting young people in the form of greater disparities which are associated with a higher incidence of social problems generally, including heavy drinking. Communities can be mobilized to initiate the organizational changes necessary for effective alcohol prevention. However, for aggregate level effects on youth drinking, strategies with demonstrated effectiveness must be implemented consistently and given sufficient time to influence drinking habits. Brief health education strategies, such as Prime for Life, may help to improve short-term attitudes towards alcohol use, but are unlikely to result in sustained behaviour change.

4 LIST OF PUBLICATIONS I. Hallgren M, Leifman H., & Andréasson S (2012). Drinking less but greater harms: could polarised drinking habits explain the divergence between consumption and harm among youth? Alcohol and Alcoholism (in press). II. III. IV. Hallgren M, Leifman H, & Andréasson, S (2012) The Swedish six community alcohol and drug prevention trial: Key findings and lessons learnt. Submitted Manuscript. Hallgren M, Källmén H, Leifman H, Sjölund T, & Andréasson S. (2008) Evaluation of an alcohol risk reduction program (PRIME for Life) in young Swedish military conscripts. Health Education, 109(2): Hallgren M, Sjölund T, Kallmén H, & Andréasson S. (2010) Modifying alcohol consumption among Swedish high school students: an efficacy trial of an alcohol risk reduction program (PRIME for Life). Health Education, 111(3): These four papers will be referred to by their roman numerals (I-IV). All articles are reprinted with the publisher s permission.

5 CONTENTS 1 INTRODUCTION Youth drinking in context Recent trends in Swedish youth alcohol consumption Recent trends in alcohol-related hospitalisations among Swedish youth Preventing alcohol-related problems in the community Relationships between drinking and harm: the total consumption model A framework for prevention: the Systems approach Evidence from previous community projects The evidence base: what works in alcohol prevention? Risk and protection: the building blocks of prevention Some words on program evaluation A brief rationale for this thesis AIMS Research questions The red thread METHOD Study 1: Polarised youth drinking The Stockholm Student Survey Statistical analyses Study II: The Swedish Six Community Alcohol and Drug Prevention Trial Brief project history Study design and interventions Prevention work in the control communities The prevention index Measures Survey participants Statistical analyses Prime for Life evaluation (studies III and IV) The intervention Military conscripts (Study III) Statistical analyses High school students (Study IV) Statistical analyses Ethical considerations RESULTS Article I polarized youth drinking Risk factors for harmful drinking Article II community based prevention of hazardous youth drinking Article III Prime for life conscript study Article IV - Prime for life high school study DISCUSSION... 45

6 5.1 The main findings Polarised youth drinking Implications Community prevention of youth alcohol problems Lessons learnt: possible explanations for the absence of positive program effects Comparisons with other prevention trials Implications Prime for life & brief education programs Implications Strengths and contribution Methodological issues Future research CONCLUSIONS ACKNOWLEDGEMENTS REFERENCES... 62

7 LIST OF FIGURES AND TABLES Figure 1: Per capita alcohol consumption among year 9 students (15-16 years) and year 11 students (aged years), Sweden, Figure 2: Percentage of young people who binge drink once per month or more, Sweden, Figure 3: Number of males per 10,000 inhabitants aged years admitted to hospital in Sweden with a primary or secondary alcohol-related diagnosis Figure 4: Number of females per 10,000 inhabitants aged years admitted to hospital in Sweden with a primary or secondary alcohol-related diagnosis Figure 5: Number of males per 10,000 inhabitants aged years admitted to hospital in Stockholm with a primary or secondary alcohol-related diagnosis Figure 6: Number of females per 10,000 inhabitants aged years admitted to hospital in Stockholm with a primary or secondary alcohol-related diagnosis Figure 7: Conceptual model of factors influencing alcohol consumption Figure 8: Overview of the research questions and study design Figure 9: The prevention index used to assess differences between trial and control community prevention efforts (2006) Figure 10: Participation and drop-out rates conscript study Figure 11: Changes in per capita alcohol consumption and alcohol related hospitalisations among Stockholm youth ages years Table 1: Examples of community trials incorporating elements of the systems approach to prevention Table 2: Specific programs included in the prevention work by the six trial communities Table 3: Variables included in the Prevention Index Table 4: Retention of participants over time, by condition and total number (conscript study) Table 5: Mean number of participants per school (m), Intracluster Correlation Coefficient (ICC) and Variance Inflation Factor (VIF) for the primary outcomes measured at baseline (high school study) Table 6: Changes in per capita alcohol consumption by percentile rank for year 9 males Table 7: Risk factors for harmful alcohol consumption (total sample) Table 8: Risk factors for harmful alcohol consumption (top 5% of drinkers)

8 LIST OF ABBREVIATIONS AUDIT BAC CAN CV DUI ESPAD MI PFL RBS SET SNIPH STAD WHO ÖPP Alcohol Use Disorders Identification Test Blood Alcohol Concentration The Swedish Council for Information on Alcohol and Drugs Coefficient of Variation Driving Under the Influence European School Survey Project on Alcohol and other Drugs (ESPAD) study Motivational Interviewing Prime for Life Responsible Beverage Service Social and Emotional Training Swedish National Institute for Public Health Stockholm Prevents Alcohol and Drug Problems World Health Organisation Örebro Prevention Project

9 1 INTRODUCTION 1.1 YOUTH DRINKING IN CONTEXT Why devote attention to youth drinking? Adolescents are consistently over-represented in alcohol-related harm statistics (WHO, 2011). Compared to other age groups, they are more likely to be harmed or seriously injured as a consequence of their drinking. It is also known that a sizable proportion of death and disability among youth is attributable to alcohol (Toumbourou et al., 2007). In addition to short-term negative outcomes, such as accidents and intoxicated aggression, evidence suggests that brain development may be adversely affected by alcohol (Lubman et al., 2007). The harmful effects that many young people experience arise partly from the amount of alcohol that they consume (consumption typically peaks in the early twenties), and from the pattern of drinking (heavy episodic drinking is more frequent among youth, and associated with serious acute harms). A recent World Health Organisation study found that out of 73 participating countries, hazardous and harmful drinking patterns, including drinking to intoxication, appear to be on the rise among adolescents (WHO, 2011). Although aggregate level data can hide important differences that exist between countries, this recent study highlights a concerning world-wide trend. The reasons for this increase are complex, but greater alcohol availability is a likely explanation. Another contributing factor could be the popularity of alcopops or alcoholic carbonate drinks, which are associated with more problematic drinking patterns, earlier onset of drinking and drunkenness (Kraus et al., 2010). On the other hand, positive developments have also been noted. The ongoing European School Survey Project on Alcohol and other Drugs (ESPAD) study, indicates that the perceived risks associated with heavy drinking among European youth have increased, and that disapproval of binge drinking among upper secondary school students has also risen (Hibell et al., 2009). These are favourable shifts, but as will be discussed, changes in attitudes do not always translate into positive changes in behaviour. From a developmental perspective, early experiences with alcohol are known to increase the risk of later alcohol use disorders (Hingson et al., 2006, Kelly et al., 2011). In particular, the age of first drinking occasion is a strong predictor of alcohol-related problems in adulthood (Pitkanen et al., 2005, Poikolainen et al., 2001, Cable and Sacker, 2008). Adolescence is a period when parents and peers have a substantial influence on behaviour and several studies have shown that adolescents who socialise with heavy drinking peers, or whose parents routinely offer them alcohol, are more likely to develop problems associated with alcohol (Ary et al., 1993, Becker and Grilo, 2006, Cable and Sacker, 2008). Moreover, recent research indicates that heavy drinking during the teen years may contribute to the development of social and health inequalities in adulthood (Hill et al., 2000, Odgers et al., 2008, Viner and Taylor, 2007). Across all age groups, alcohol is a causal factor in more than 60 major diseases and injuries and it s consumption results in approximately 2.5 million deaths each year (WHO, 2011). About 4 per cent of all deaths worldwide are attributable to alcohol, where it is the leading risk factor in the world for deaths among young males. Four and a half per cent of the global burden of disease and injury is attributable to alcohol. A recent Swedish government report estimates that the economic cost of alcohol consumption in Sweden is around 66 billion SEK annually (SOU, 2011). 1

10 Litres of puer alcohol In sum, both the magnitude and frequency of alcohol-related harms among youth offer compelling reasons to better understand the nature of youth drinking and what can be done to prevent unnecessary injury and death. This thesis contributes to the field with recent data from the Swedish context Recent trends in Swedish youth alcohol consumption The Swedish Council for Information on Alcohol and Drugs (CAN) has conducted annual public school surveys of alcohol consumption among year 9 students (aged years) since 1971, and year 11 students (aged years) since The anonymous self-report surveys measure the quantity and frequency of different types of alcohol, enabling the calculation of a yearly estimate of total consumption. As the survey is completed during class time, response rates have consistently been high. In 2011, for example, year 9 students, and year 11 student participated in the survey, with response rates of 83 and 81 per cent, respectively (Henriksson and Leifman, 2011). Fifty-five per cent of year 9 boys and 59 per cent of year 9 girls indicated they had drunk alcohol at least once during the previous 12 months. The figures for year 11 boys and girls were 83 and 84 per cent, respectively; the lowest rates that have been recorded for both age groups since the survey began. For year 9 students, consumption peaked in 2000 before reducing steadily until Data for year 11 students is only available from 2004, where we also see a steady decline in consumption over the past seven years, predominantly among males. These changes are illustrated in Figure 1, below Yr 9 Males Yr 9 Females Yr 11 Males Yr 11 Females Figure 1: Per capita alcohol consumption among year 9 students (15-16 years) and year 11 students (aged years) in Sweden, Source: The Swedish Council for Information on Alcohol and Drugs (CAN). Drug Trends in Sweden, 2011.Report nr. 130, Stockholm In 2011, the estimated per capita consumption for year 9 females was 1.8 liters of pure alcohol and 2.2 liters for boys; again, the lowest recorded levels since 1996 and 1988, respectively. Per capita consumption also reduced among year 11 boys from 6.8 liters in 2004 to 5.5 liters in The trend among year 11 females was more stable, dropping from 3.9 to 3.4 liters during the same period. The reduction in per capita consumption has been driven primarily by an increasing number of young people who abstain from alcohol completely. However, reductions are also seen when only the 2

11 Binge drink once per month or more (%) alcohol consumers are examined, except among Year 11 females, where a small rise between 2004 and 2011 has been observed (Henriksson and Leifman, 2011). Heavy episodic drinking, or binge drinking, is more prevalent among youth and tends to decline with age; a trend observed in most countries worldwide (WHO, 2011). Binge drinking has been a public health concern in Sweden for decades because it is a pattern of consumption strongly associated with acute harms, such as motor vehicle accidents, violence, and acute alcohol intoxication (Toumbourou et al., 2009, Rehm et al., 2009). Due to this association, recent trends in binge drinking are of great interest. Figure 2 illustrates the steady decline in heavy episodic drinking among Swedish youth over the past decade, a trend similar to the total consumption changes shown in Figure Yr 9 Males Yr 9 Females Yr 11 Males Yr 11 Females Figure 2: Percentage of young people who binge drink once per month or more, Sweden, Source: The Swedish Council for Information on Alcohol and Drugs (CAN). Drug Trends in Sweden, Report nr. 130, Stockholm Binge drinking among Year 11 females has increased slightly, although the trend has reversed over the past three years. For the first time since 1971, Year 9 males report binge drinking less frequently than Year 9 females. This is interesting to observe because across all age groups (and in most countries worldwide) males typically drink more alcohol than females (Babor et al., 2010). In terms of beverage preferences, males in both school years continue to prefer strong beer and spirits, whereas females prefer blended drinks (now the preferred choice among Year 11 females) and spirits. Overall, males increasingly prefer to drink strong beer, and females increasingly prefer blended or mixed drinks. There has been a recent trend towards lower consumption of spirits among Year 11 males and females (Henriksson and Leifman, 2011). 3

12 1.1.2 Recent trends in alcohol-related hospitalisations among Swedish youth Young people can experience a range of harmful consequences when they drink alcohol. One of the more serious outcomes is hospitalisation due to acute intoxication or alcohol poisoning, which in Sweden accounts for the majority of all alcohol-related hospital admissions involving young people (Valdatabasen., 2010). All public hospitals in Sweden are required to provide annual data on the number of people admitted to hospital with an alcohol-related diagnosis and this information is recorded by CAN. Between 2000 and 2010, the total number of youth aged (year 9) and (year 11) admitted to hospital with a primary or secondary alcohol-related diagnosis increased from 1,078 to 1,562; a real increase of 5.7 admissions per 10,000 youth. As shown in Figures 3 and 4, the rate of increase appears to be driven mainly by adolescents aged years, who have recently overtaken their younger peers in terms of annual alcohol-related hospital admissions nationally. The rise in admissions due to acute intoxication or poisoning has been particularly striking in Sweden s capital, Stockholm (Figures 5 and 6). Between 2000 and 2010, the number of admissions increased by 17 per cent among year-olds, and 29 per cent among year-olds (Ahacic and Thakker, 2010, Valdatabasen., 2010). These figures represent unique cases (as opposed to repeat admissions) per 10,000 inhabitants with a diagnosis of acute intoxication and/or alcohol poisoning upon admission. Compared to the national figures for Sweden, the main differences are firstly, a higher proportion of admissions in Stockholm, and secondly, a clear increase in admissions among year olds. Nationally, the trend for year old adolescents has been more stable over time, with a recent decline. The largest increase has been among females aged years in Stockholm. These young women are the only group to show signs of increasing binge drinking, and a strong preference for mixed drinks with high alcohol content (CAN, 2011). The marked drop in all admissions seen in 2003 is most likely the result of a change in the admission recording procedures in one of the major hospitals that year. It should be noted that the higher hospital admission rate in Stockholm compared to the rest of Sweden could be due to underlying differences in drinking patterns between urban and rural adolescents, or equally, they may reflect differences in service access or treatment opportunities, which could be higher in the country s capital city. Together, this data shows a divergence between alcohol consumption, which has reduced over the past decade, and alcohol-related hospitalisations, which have risen an unexpected trend that deserves explanation and provides the starting point for a detailed analysis of drinking trends in Study 1. 4

13 Figure 3: Number of males per 10,000 inhabitants aged years admitted to hospital in Sweden with a primary or secondary alcohol- related diagnosis (acute alcohol intoxication or poisoning, ICD codes F100 or T51). Source: Socialstyrelsen, Figure 4: Number of females per 10,000 inhabitants aged years admitted to hospital in Sweden with a primary or secondary alcohol- related diagnosis (acute alcohol intoxication or poisoning, ICD codes F100 or T51). Source: Socialstyrelsen,

14 Figure 5: Number of males per 10,000 inhabitants aged years admitted to hospital in Stockholm with a primary or secondary alcohol- related diagnosis (acute alcohol intoxication or poisoning, ICD codes F100 or T51). Source: Valdatabasen, Figure 6: Number of females per 10,000 inhabitants aged years admitted to hospital in Stockholm with a primary or secondary alcohol- related diagnosis (acute alcohol intoxication or poisoning, ICD codes F100 or T51). Source: Valdatabasen,

15 1.2 PREVENTING ALCOHOL-RELATED PROBLEMS IN THE COMMUNITY Early and popular views regarded high-risk individuals as the main source of alcoholrelated health problems. Education and information was the dominant prevention strategy, while the underlying social and community mechanisms responsible for alcohol-related harms were largely overlooked (Room, 1997, Gruenewald, 2011). There was a keen focus on alcoholism, and a widely held view that heavy problematic drinkers those most visible in society were the source of most alcohol-related harmful effects. This view stemmed from the Medical Model which has an individualistic perspective on the nature of addiction, and which dominated thinking until at least the 1960 s (Edwards, 1978). To change problematic drinking behaviour, one needed to modify the problematic individual responsible for the harmful alcohol use. In the 1960 s and 70 s alcohol researchers began to question this focus. It was noted that there was no threshold at which one suddenly became a significant risk for alcohol-related problems and that there was some risk for harm at consumption levels below that associated with alcoholism (Stockwell et al., 1997). Recent prevention research and behaviour change theory has also shifted this perspective substantially (Birckmayer et al., 2004, Petraitis et al., 1995, Foxcroft and Tsertsvadze, 2011a). We now know, for example, that the greatest harms from alcohol arise not from a limited number of severely problematic drinkers, but from the larger group of heavy drinkers with less severe problems, a scenario referred to as the prevention paradox.(kreitman, 1986). Most investigators agree that a highly effective way to reduce alcohol problems is to target whole populations; not only high-risk individuals. Policies which regard alcohol as a public health issue and a subject for comprehensive regulation have been uncommon outside Sweden and the Nordic countries. Since 1995 when Sweden joined the European Union, however, there has been a decline in Swedish alcohol control policy, including those interventions which have the greatest potential for curtailing alcohol-related problems. Between 1995 and 2004, per capita consumption increased by around 30 per cent in Sweden, a development which prompted the adoption of a national alcohol action plan in 2000 (and revised in 2005). Among other things, a stronger emphasis was given to prevention work in local communities (the focus of Study II in this thesis). The sections which follow summarise four topics that are central to Swedish alcohol prevention policy: (1) the relationship between consumption and alcohol-related harmful effects; (2) the systems approach to community prevention; (3) the evidence base for prevention, and lastly (4) the risk/protection model. As will be discussed, to some degree, the systems model and other prevention strategies based on risk and protection theory, overlap. 7

16 1.2.1 Relationships between drinking and harm: the total consumption model Studies from several countries demonstrate that alcohol consumption is very unevenly distributed in a population; most alcohol is drunk by a relatively small minority of drinkers (Babor et al., 2010, Norstrom and Ramstedt, 2005). There is also a strong relationship between the per capita alcohol consumption, the prevalence of heavy drinking, and alcohol-related problems. This relationship forms the basis of the total consumption model, which has been influential in Sweden. In their classic article The population mean predicts the number of deviant individuals, Rose and Day (1990) demonstrated that for various health risk indicators there is a strong association between the population mean and the prevalence of problems. They concluded that the distributions of health related characteristics move up and down as a whole: the frequency of cases can be understood only in the context of a population s characteristics (Rose and Day, 1990). With respect to alcohol consumption, Ole-Jörgen Skog has argued that changes in per capita consumption tend to influence all levels of consumption concurrently, including heavy drinking (Skog, 1985). Consequently, when mean consumption increases or decreases, the proportion of heavy drinkers should change accordingly (Skog and Rossow, 2006); a phenomenon driven mainly by strong social influences on drinking behaviour within cultures. Consistent with Skog s theory and the total consumption model, current Swedish alcohol policies aim to reduce population level drinking and associated harms through restrictions on the availability of alcohol through (among other things) a retail monopoly, age checks at the point of alcohol purchase, and regulations over trading hours. Historically, increases in alcohol availability in Sweden have been associated with increased per capita consumption, and more alcohol-related mortality and morbidity, which support the total consumption model (Holder, 2000b, Andreasson et al., 2006, Norstrom and Ramstedt, 2005). For example, increases in consumption and alcohol-related harms were observed shortly after Sweden joined the European Union in 1995 when traditional protections were eroded through increased cross-border trade and lower excise duties (Holder, 2000b). While a relationship between total consumption, heavy drinking and alcohol-related harms has been observed, exceptions to this general association have been noted which may have consequences for Swedish alcohol policy. For example, a recent study examining changes in alcohol-related harms in northern and southern Sweden after increased alcohol imports from Denmark, failed to show a uniform increase in harms associated with more alcohol availability (Gustafsson and Ramstedt, 2011). Similarly, Study 1 in this thesis, Drinking less but greater harm, also highlights an exception to this relationship A framework for prevention: the Systems approach There are powerful advantages to population level prevention of alcohol problems. This type of prevention attempts to remove or modify the underlying cause of the problem and has considerable potential to bring about change due to the large number of individuals involved (Loxley et al., 2004). Harold Holder s systems model of community based alcohol prevention has been influential in Sweden, and guided the development of the Swedish Six Community Alcohol and Drug Prevention Trial (Study 8

17 II in this thesis). His theoretical model, described in Alcohol and the Community: A systems approach to prevention (Holder, 1997), regards the community and it s multiple sub-systems as the main target for intervention efforts at the local level. All communities, he argues, consist of individuals and entities that influence each other in a socio-cultural-political-economic context (Holder 1997, p. 12). To have maximum effect, prevention efforts need to be directed towards as many system-wide structures and processes as possible. Uni-dimensional strategies, such as education about the harmful effects of alcohol, are unlikely to be effective, unless other parts of the system are primed (or mobilised) to respond to such messages. Important sub-systems which can influence alcohol problems within a community include drinking patterns, alcohol availability, enforcement efforts, sanctions and social norms. An overarching aim of the systems approach is to achieve prevention rather than treatment of existing alcohol problems in the community. To provide an example of the model s application, a local high school could be seen as a sub-system in which the behaviour of students is influenced by national alcohol regulations (availability, price, age-restrictions), the physical environment of local drinking establishments (location, crowding, noise), the behaviour of bar staff (responsible beverage service practices), public opinion regarding drunkenness (parental and peer influences), and the scrutiny of local police (enforcement). A systems approach to prevention aims to identify these underlying community-level risk factors, and to modify them in order to reduce problematic drinking. The systems model advocates the use of both supply measures, which limit access to alcohol, and demand measures, which reduce individual demand for alcohol. Many of these causal factors have a bi-directional influence, as shown in Figure 7, below. This model, adapted from Birchmayer and Holder et al (2004), illustrated the main areas targeted by alcohol prevention at the community level, namely: availability, enforcement, social norms and alcohol promotion. The model recognises the association between availability, per capita consumption and alcohol-related harms. A community system perspective calls for approaches that go beyond education, screening, and other individually focussed programs, and instead attempts to change the environment (broadly defined) related to risky drinking behaviour. Holder and others have noted the importance of creating effective partnerships between researchers who develop science-based interventions, and practitioners who implement and sustain such interventions locally. This emphasis, and the need to modify risky drinking environments (as opposed to risky individuals), has been a central focus of successful prevention efforts in Australia, Canada and New Zealand recently (Livingston, 2008, Stockwell et al., 2011, Connor et al., 2011, Homel et al., 2004) The implementation of a systems approach largely involves legislative change and enforcement. Integrating research into the evaluation design can be costly, which possibly explains why only a small number of interventions based on this model have been implemented and evaluated worldwide. 9

18 The focus of community based alcohol prevention Enforcement Social norms Parents and peers Availability: Price Retail structure Trading hours Outlet density Age restrictions Responsible beverage service Promotion Alcohol and health industry Per-capita alcohol consumption Heavy episodic drinking Alcohol-related harms Figure 7: Conceptual model of factors influencing alcohol consumption (Adapted from Birckmayer et al (2004) Evidence from previous community projects Support for community prevention based on the systems model - or variations of this model - has grown as a small but increasing number of trials have demonstrated positive effects. Four community prevention projects undertaken in different parts of the world are described below and a summary of recent trials is set out in Table 1. This is not an exhaustive list, but it does illustrate the main features of different prevention programs. 10

19 In the United States, a five year alcohol prevention project (the Three Communities Trial ) was conducted between 1992 and 1996 to determine the effect of environmental prevention strategies on alcohol-related injury in three intervention communities (Holder et al., 2000). The interventions included community mobilisation, responsible beverage service, age checks, increased local enforcement of drink-driving laws, and zoning to limit access to alcohol. By the end of the trial, self-reported alcohol consumption had declined by 6 per cent; the frequency of having had too much to drink reduced by 49 per cent; drink-driving reduced by 51 per cent; and night-time vehicle crashed declined by 10 per cent. In addition, assault injuries observed in emergency departments declined by 43 per cent in the intervention communities. Also in the US, the Communities That Care project ( ) aimed to reduce adolescent alcohol and drug use and delinquent behaviour communitywide (Hawkins et al., 2009). Twenty-four small towns in seven states were randomly assigned to control or the intervention condition. The participants were 4407 youths aged years. The intervention involved the collection of epidemiological data to identify elevated risk factors and depressed protective factors in the community and the implementation of tested programs to address the community s specific needs. Unlike the Three Communities Trial (Holder et al., 2000), this project did not focus exclusively on the prevention of alcohol use, but on reducing risk factors that predict early initiation and use among youth, in addition to other health-risking behaviours such as delinquency. Also unlike other prevention trials in the US (e.g., Project Northland, Communities Mobilising for Change on Alcohol), environmental risk factors such as venue opening hours, age-checks, and regulatory enforcement, were not targeted. Results indicated that alcohol use, cigarette smoking and delinquent behaviour were significantly lower in the trial communities than in the control areas for students in grades 3 through 8 at follow-up in Binge drinking during the last two weeks, and alcohol consumption during the last 30 days both reduced significantly during the project. A community intervention project in the Northern Territory, Australia, aimed to reduce higher levels of alcohol-related harm to national levels using a range of strategies, including a levy on alcoholic beverages with more than 3 per cent alcohol to fund education, increased controls on alcohol availability, and expanded treatment and rehabilitation services (Chikritzhs et al., 2005). The intervention led to a significant preferential reduction in acute alcohol-related deaths and to a non-significant reduction in chronic, alcohol-related deaths in the Northern Territory compared to the control areas. Finally in Trelleborg, southern Sweden, a three-year community intervention trial was conducted targeting youth drinking (Stafstrom et al., 2006). The interventions included the adoption of a community and school policy and action plan on alcohol and drug management; increased Police inspections of grocery and convenience stores where black market alcohol could potentially be sold; the introduction of an evidence based curriculum on alcohol and drugs in schools; information and support for parents, and the use of mass media to boost knowledge about alcohol related harms. Results from the trial were positive and included a 20 per cent decrease in the proportion of alcohol consumers (compared to a 5 and 1 per cent increase in two control areas, and a 5 per cent increase nationally). Similar trends for excessive drinking and heavy episodic drinking during the last month were also observed. 11

20 In a recent Cochrane report, David Foxcroft and collaborators identified and systematically reviewed 20 methodologically sound, multi-component alcohol prevention trials targeting young people (Foxcroft and Tsertsvadze, 2011a). Twelve of the 20 trials reported statistically significant effects across a range of outcomes in the short and long-term. Six trials, however, found no effects on youth alcohol consumption or related harms. The authors concluded that, overall, current evidence supports the effectiveness of some multi-component programs targeting young people, with effect sizes that are often small, but potentially important. The authors noted that more needs to be understood about the content and context effects of community trials. In other words, trials need to be evaluated in different contexts, and they should include a detailed description of the various program components, and (where possible) assessments of their relative impact on the outcomes measured (Foxcroft and Tsertsvadze, 2011a). Also relevant here is a recent Norwegian report which highlights the utility of mixed methods in the evaluation of community trials. The authors suggest that qualitative methods can greatly assist the interpretation of quantitative findings (Rossow and Baklien, 2011). The programs listed in Table 1 include examples of recent successful community interventions to reduce alcohol consumption (e.g. Communities That Care; the Trelleborg Project), and programs that did not result in significant improvements during the intervention period (e.g. DANTE Victoria; Project Northland, Chicago), illustrating that not all community trials are successful. This also reinforces the importance of evaluation and the need to explain negative findings when they arise. As will be discussed in Study II, a myriad of factors can influence the success of community based interventions, including the extent to which local communities are actively engaged in the project, the choice of intervention strategies, the intervention dose and fidelity (i.e. if the program was implemented as intended), the study design and evaluation method, and policy changes during the trial period. 12

21 Table 1: Examples of community trials incorporating elements of the systems approach to prevention Project Alcohol: less is better project (Bagnardi et al., 2011) Dealing with Alcohol Related Problems in the Night-time Economy (DANTE) (Miller et al., 2011) Communities That Care (Hawkins et al., 2009) Project Northland Chicago: (Komro et al., 2008) Objective To reduce community level alcohol consumption; and improve community attitudes towards alcohol use. To reduce alcohol related emergency department admissions to hospital To reduce adolescent alcohol and drug use and delinquent behaviour communitywide To reduce total and risky drinking among urban youth, including mediating risk factors Location & Design Italy, Controlled intervention trial with pre-post intervention comparisons Geelong, Australia, Pre-post intervention study design Seven states in the USA, Randomised controlled trial Chicago Randomised controlled trial Population Ten trial and eight matched control communities; in total 123,235 inhabitants (trial only) 217,000 people living in the City of Geelong, Australia Forty-one communities within the seven states participated. A panel of 4407 fifth grade students was surveyed annually 61 schools and surrounding areas (baseline), and 5698 students in grades 5-8 Interventions Extensive community mobilisation Educating and sensitising the population on alcohol problems via schools and within local communities Venue based prevention: ID scanners for age checks, police enforcement, local accords Media: safer drinking promotions School & community based youth focussed programs Family focussed programs Classroom curricula Parental involvement Peer leadership Youth extra-curricula activities Community organising Outcomes Significant reductions in self-reported alcohol consumption in the trial communities Overall, the intervention was not associated with reduced alcohol related hospital admissions Significant reduction in the incidence of alcohol and drug use among students in the trial communities only No significant changes in the intervention communities A trend towards reduced alcohol access by youth 13

22 Table 1 continued: Project The Trelleborg Project (Stafstrom et al., 2006) Over serving at licensed premises in Stockholm (Wallin et al., 2005, Wallin et al., 2002) Communities Mobilising for Change on Alcohol (CMCA) (Wagenaar et al., 2000) Three Communities Trial (Holder et al., 2000) Objective To reduce harmful drinking patterns among adolescents, delaying age of first onset drinking, improving attitudes towards alcohol To reduce alcohol service to intoxicated patrons To reduce violence in the project area, Stockholm To implement policies and regulations aimed at reducing youth access to alcohol To improve the public health of the trail communities To reduce alcohol-related accidental injuries and deaths Location & Design Trelleborg, Sweden, Pre-post intervention comparisons with national and regional drinking trends Stockholm, Quasi-experimental design Minnesota, Wisconsin, USA, Group randomised trial California, USA, Matched intervention and control group design Population 39,000 adults and youth in the Trelleborg region Central Stockholm (~300,000 inhabitants), with Södermalm as comparison area (~100,000 inhabitants) 7 communities with an average population of 21,000 inhabitants 8 control sites Three trial communities in northern and southern California with a population of approx. 100,000 each Interventions Police enforcement School based curriculum Parent information on youth drinking Media interventions Community mobilisation Responsible Beverage Service training Enforcement 2.5 year community organising initiative involving extensive mobilisation and media/information dissemination in the trial communities Five components targeting: Local mobilisation Responsible Beverage Service Drinking and driving Under age drinking Access to alcohol Outcomes Decreased harmful drinking behaviour prepost intervention Increased abstention rates Violent crimes decreased significantly by 29% in the intervention areas Significant declines in the trial communities for DUI arrests among yearolds DUI arrests and disorderly conduct violations among year-olds approached significance Reduction in night-time injury crashes by 56 per 100,000 adult population per year Reduced drink-driving crashes by 67 per 100,000 adult population 14

23 1.2.4 The evidence base: what works in alcohol prevention? Any discussion about youth alcohol consumption must consider the evidence base for prevention. Fortunately, a great deal is known about what works in alcohol prevention, and much of this knowledge is summarised in the book Alcohol: No Ordinary Commodity (Babor et al., 2010). Recent and comprehensive reviews have also contributed to what is known about effective prevention (Foxcroft et al., 2011, Anderson et al., 2009a, Stockwell et al., 2003). Many of the strategies set out below target whole populations (for example, price and availability restrictions), while others target school aged youth (for example; peer resilience programs, school based education). There is ongoing discussion in the literature regarding what constitutes the optimal balance between targeted versus community-wide prevention. On balance, the consensus is that a combination of both population level strategies and targeted interventions for high-risk youth is likely to achieve the greatest benefit (Babor et al., 2010, Foxcroft and Tsertsvadze, 2011a, Toumbourou et al., 2007) Availability Most investigators agree that reducing alcohol availability across multiple domains (community, home, peers) is the most effective strategy to reduce harmful drinking and alcohol-related problems (Babor et al., 2010, Anderson et al., 2009a). Studies have also shown that greater access to alcohol increases the odds for adolescent binge drinking, drunkenness, and belonging to a higher consumption trajectory group (Danielsson et al., 2010, Patrick and Schulenberg, 2010). A recent study examining the relationship between alcohol control policies and adolescent alcohol use in 26 countries found that more stringent policies, particularly those affecting availability, were associated with lower prevalence and frequency of adolescent drinking and age of first alcohol use (Paschall et al., 2009). Reducing access to alcohol can be achieved in several ways: reducing the density of alcohol outlets in the community (Stockwell et al., 2011), limiting trading hours (Rossow and Norstrom, 2012), age restrictions at the point of sale (Wagenaar and Toomey, 2002), and Responsible Beverage Service (RBS) practices (Wallin and Andreasson, 2004, Livingston, 2008). It can also be achieved through government control of alcohol distribution and sales. One form of government control is the retail monopoly system which exists in Sweden, Systembolaget. A US study exploring associations between state retail alcohol monopolies, underage drinking and alcoholimpaired driving deaths, found that monopolies over both wine and spirits were associated with larger consumption reductions than monopolies over spirits only. Lower consumption rates, in turn, were associated with a 9.3 per cent lower alcoholimpaired driving death rate (Miller et al., 2006). Similarly, recent Swedish studies have demonstrated critical links between alcohol availability and alcohol-related mortality and morbidity (Andreasson et al., 2006, Norstrom et al., 2010) Price One of the most effective strategies for reducing consumption at the population level is through increasing alcohol prices. A recent review of 112 studies on the effects of alcohol tax affirmed that when alcohol taxes increase, drinking goes down including problem drinking among adolescents (Wagenaar et al., 2009). Although price restrictions can have beneficial effects from a public health perspective, the strategy is not favored by most countries due to the detrimental impact of such policies on the highly competitive alcohol industry. 15

24 Drinking environments In Sweden, the legal age for purchasing alcohol from bars, restaurants and clubs is 18 years. Many adolescents choose to drink alcohol in or near licensed venues, where going out is seen as a rite of passage and where experimenting with alcohol is common. Recent prevention research has shown that drinking environments can influence drinking behaviour and associated violence (Graham and Homel, 2008, Foxcroft and Tsertsvadze, 2011a, Wallin et al., 2002). High-risk drinking venues characterised by over-crowding, poor staff training and patron discomfort have been linked to higher rates of alcohol-related problems (Homel et al., 2004). In Australia, Ross Homel and collaborators have identified several venue-level factors associated with harmful drinking (Homel et al., 2004, Graham and Homel, 2008). These include over-crowding, poorly trained staff, heavily intoxicated patrons, inadequate public transport, late closing hours, and chap drink specials. In Sweden and elsewhere, several studies have demonstrated that modifying these risk factors can lead to significant reductions in alcohol related violence, with large cost-savings for the community (Graham et al., 2005, Wallin et al., 2002, Homel et al., 2004, Mansdotter et al., 2007) Drink-driving countermeasures Alcohol consumption is associated with a higher incidence of traffic accidents worldwide (Rehm et al., 2009). A recent New Zealand study found that the rate of road traffic injuries and the involvement of alcohol peaks during late adolescence, as does the proportion of all road traffic injuries that are caused by other people drinking (Connor and Casswell, 2009). Setting maximum blood alcohol concentrations for drivers and enforcing these with random breath testing can reduce alcohol-related motor-vehicle crashes by 20 per cent (WHO, 2011). Moreover, setting lower BACs for younger drivers can reduce alcohol-related crashes among this population by between 4 and 24 per cent (Shults et al., 2004) Alcohol promotion Research has shown that the level of alcohol advertising in a community is associated with alcohol-related problems, including road fatalities (Smith and Foxcroft, 2009). The strongest evidence for the association comes from longitudinal studies that have shown an effect of various forms of alcohol marketing on the initiation of youth drinking, and on riskier patterns of youth drinking (Anderson et al., 2009b). Historically, alcohol advertisements in Sweden have been prohibited. However, marketing is allowed for beverages identified as class 1 (for example, light beer), and since 2005, newspaper advertisements for alcohol were permitted under EU directives. Despite the finding that a general association exists between the level of advertising in a community and alcohol-related harms, a recent systematic review of advertising bans found inconclusive results, mainly due to methodological limitations (Booth et al., 2008). Other studies have emphasized the link between alcohol promotion and drinking levels among adolescents. A recent Australian study of 1113 adolescents aged years found that exposure to alcohol advertisements was strongly associated with drinking patterns (Jones and Magee, 2011). Similarly, a recent review of prospective cohort studies suggests that there is an association between exposure to alcohol advertising or promotional activity and subsequent alcohol consumption in young people (Smith and Foxcroft, 2009). 16